An animal is just its body and its place in the chain
For a long time, a creature in The Long Watch was a pile of numbers. How much it ate, how long it lived, how big a litter it raised, how crowded a patch could get before it starved — each one dialed in by hand, by feel, one species at a time. This is the story of how we stopped doing that, and let an animal become something the world could simply describe.
The problem with hand-tuning a creature is that it doesn’t scale. A world meant to hold sixty or seventy kinds of animal can’t be hand-calibrated sixty or seventy times over — and every time you add one, the delicate balance you struck for the others has to be re-struck. We had felt this already on the small roster, where finding a setting that held even four creatures together took a hundred parallel worlds to measure. Doing that for every new arrival was never going to work.
Two facts, and the rest is derived
So we flipped the question. Instead of authoring all those numbers, you now describe a creature by just two things: how heavy its body is, and where it sits in the food chain. A body mass, and a place — a level from one to five, and a role like grazer, predator, apex, or scavenger. From those two facts, the world works out the rest on its own.
It can do that because real animals already obey these rules, and we borrowed them. A heavier body lives slower and longer. It ranges over more ground. It breeds more slowly and raises fewer young at a time. It burns energy at a different rate relative to its size. These are some of the oldest, broadest patterns in biology — the way a mouse and an elk differ isn’t arbitrary, it tracks their size — and once you encode them, a single number, body weight, fans out into a whole life-rhythm: metabolism, lifespan, litter size, the size of a home range.
A species stops being a calibration and becomes a description.
That sentence is the whole point. Before, adding an animal meant sitting down with thirty knobs and turning each one until the creature felt right. Now it means writing down what the animal is — its weight, its rank, what it eats — and letting those few facts imply everything downstream. The designer’s job moves from tuning to describing, which is the job we actually wanted all along.
How many can live here, decided by the ground
The same shift reached the hardest number of all: how many of a creature a place can hold before they start to starve. That used to be a fixed figure written next to each species. Now it’s computed live, from the state of the world itself.
For the plant-eaters, the ceiling comes from the ground beneath them — how much forage is actually growing there, how fertile the soil is, how well the spot matches the biomes that species favours. A lush meadow can feed many; a thin, cold patch can feed only a few. Nothing about that limit is authored. It rises and falls with the living world, so the same kind of animal settles at a different density on rich ground than on poor.
The hunters near the top of the chain needed something different, and the reason is quietly interesting: a predator’s food doesn’t visibly thin out as the predator multiplies. Grass gets eaten down and you can see the patch run short; prey scattered across a whole landscape doesn’t announce its scarcity the same way. So a predator left to a forage-style brake would just keep climbing. Instead, the limit on a ranging hunter falls out of its body weight as a question of territory: a heavier hunter needs more ground to itself, so fewer of them fit per square of the world. The bigger the animal, the sparser it has to live — which is, again, exactly how it works outside the game.
The proof was a creature that was nothing but a description
A principle like this is only worth as much as its first honest test. The principle is that the world can derive a whole animal from a description. The proof had to be one animal that was only a description — no hand-written life-rhythm, no hand-set crowding cap, none of the manual override knobs the earlier creatures leaned on.
That animal is the stoat. It’s a small, solitary, ranging hunter — the lightest creature in the world, about a quarter of a kilogram, against a fox at roughly six. We authored it as pure data: its weight, its place in the chain, what it eats, and nothing else. Where the fox still carried several hand overrides, the stoat carried zero.
The detail that made it a real test, and not a creature quietly engineered to pass, is that the stoat’s weight sat outside the small set of animals the scaling laws had been fitted against. It was a genuine out-of-sample case — lighter than anything the rules had seen. If the derivation only worked for the creatures we’d used to build it, the stoat would expose that.
It didn’t. In a short run of a world the stoat established itself, hunted rabbits as part of the wider loop, grew toward the territorial ceiling its own body weight implied, and was held there by crowding rather than running away unchecked. Seeded with a small founding group, it climbed to a couple of hundred and then levelled off, right where its size said it should. The world had worked it out from its weight and its place in the chain, and the animal behaved.
Alongside the stoat we wrote down the recipe — a plain authoring contract — so the remaining fifty-odd species can be added the same way: as descriptions, not as code. It’s the same payoff the plant side reached first, when we grew the flora from two species to seven without writing a new rule. A foundation built broad enough lets new life arrive almost for free.
One honest caveat
We proved one branch end-to-end, not both. The ranging-hunter path — the territorial-ceiling side — is the part the stoat exercised and confirmed. The grazer path, where a population is bounded by the ground that feeds it, leans on a part of the food-web model that isn’t finished, so we’re naming it as the next thing to prove rather than claiming it works. Proving one path is not the same as proving both.
There’s a discipline underneath all of this that we keep returning to, and it’s worth stating plainly. A creature’s weight and its place in the chain are chosen from real ecology — what the animal actually is — and the population that results is then checked. It is never back-solved from a number we wanted to see. If you reverse that — pick the population you’d like and work backwards to a body that produces it — you’re no longer describing an animal, you’re just hiding the old hand-tuning behind a new spreadsheet. The whole value of the inversion is that the world is the one doing the arithmetic, honestly, from facts that mean something.
What we’re left with is an animal that is, in the end, a very short story: how heavy it is, and where it stands among the things it eats and the things that eat it. Everything else — how long it lives, how fast it breeds, how thickly it can crowd a hillside — the world reads off those two facts, the same way it always has for the real animals we borrowed the rules from.
